From Synthetic Gene Networks to Artificial Tissues – Syne2Arti

Synthetic biology, or bioengineering, aims at designing and constructing in vivo biological systems that performs novel, useful tasks. This is achieved by reingeneering existing systems.In contrast to traditional biotechnology, the focus is on developing biological or computational tools that help with an efficient construction of new systems. In this project, we focus in applications of synthetic biology for tissue engineering. Because engineered tissues might supplement or even completely replace defective tissues in patients, their development is therapeutically highly relevant. This requires firstly to reprogram cell growth, differentiation and death, which necessitates extensive modifications of intracellular networks, and secondly, to engineer cell-cell communications so as to coordinate the development of the tissue at the cell population level. This task is delicate since, on one hand, manipulations of intracellular molecular interactions can affect cell-cell and cell-substrate interactions interfering both with tissue architecture and function, and, on the other hand, changes in tissue architecture and function can feed back to the control of gene expression and post-transcriptional modifications. Consequently, the global behavior of engineered tissues emerges from local interactions between extensively modified cells, and this multi-scale aspect needs to be taken into account to successfully engineer such tissues. Unfortunately, because of this multi-scale aspect, the numerical simulation of the (monolithic) models developed to assist the tissue design is not computationally tractable: one has to analyze large differential equation systems (up to 10^6 ODEs) with uncertain parameters defined by probability distributions (random ODEs). Besides large computational resources, such analyses require novel methodological developments for multi-scale numerical simulations.

To deal with this problem, we propose to use of abstraction. More precisely, i) we compute an abstract representation of the original random ODE system in the form of a continuous time stochastic automaton, and ii) we use this abstract representation in place of the original ODE model for the numerical simulation of individual cell based tissue models. The abstract system is defined such that its behaviors are approximately equivalent to the ones of the original system (approximate behavioral equivalence). Three major tasks can be identified. The first one is to develop a theoretical framework to formally define this notion of approximate behavioral equivalence of random processes, and to provide algorithms and tools to compute abstractions of an original random ODE model and assess their quality. The second task is to apply this framework to the computation of an abstract representation of the complex intracellular network of engineered stem cell that are genetically modified to create an insulin-producing tissue. The last task is to develop an individual cell based model of monolayer stem and pancreatic (beta) cell tissues, and to extend it with the abstract representation of the engineered cells computed previously. The resulting model can then be used to predict the tissue development and suggest potential system improvements.